Data Gurus to Hillary: The Numbers Didn’t Lose, You Did
Democratic data experts said Thursday there was a zero percent chance that Hillary Clinton lost because the Democratic National Committee gave her bad data.
Depending on who is providing the reflection, Hillary Clinton lost the 2016 election because of Russian interference, former FBI Director James Comey’s unexpected October announcement, a surge in white-working class support for now-president Donald Trump, the media coverage of Clinton’s use of a private email server, her lack of attention to the Blue Wall states or a combination of all the above. And none of these would be necessarily wrong.
But a new excuse Clinton provided on Wednesday drew the ire of Democratic party operatives whose job it is to assist candidates with data for campaigns. They took this one personally as it called their jobs into question and passed the buck of responsibility onto data they deemed to be reliable.
“I set up my campaign and we have our own data operation. I get the nomination. So I’m now the nominee of the Democratic Party. I inherit nothing from the Democratic Party,” Clinton said during an interview at Recode’s Code Conference on Wednesday. “I mean it was bankrupt, it was on the verge of insolvency, its data was mediocre to poor, nonexistent, wrong. I had to inject money into it — the DNC — to keep it going.”
To say the least, this is not at all how data gurus saw it.
“Forgive the analogy,” John Hagner, currently a partner at Clarity Campaign Labs told The Daily Beast. “The DNC is farming and what the campaign does is cooking. It’s hard to blame the farmer if the souffle folds.”
Hagner, who previously worked as the National Field Director for the Democratic Senatorial Campaign Committee, said that his company was primarily focused on governor’s races in the 2016 cycle. And he pointed to one particular victory as proof that the DNC data itself was not the problem: the narrow election of Governor Roy Cooper in North Carolina, a state that Clinton lost.
“We worked out of the same pool the Clinton folks did,” Hagner continued. “The Clinton folks did the targeting themselves. The Clinton campaign did a lot right. The world ended up working out differently than they hoped.”
For many, the Clinton comments came as a surprise—and not necessarily a welcome one.
Andrew Therriault, who previously worked as the DNC’s director of data science and left halfway through 2016, called the remarks “fucking bullshit” in a series of tweets last night that have since been deleted.
“DNC data folks: today’s accusations are fucking bullshit, and I hope you understand the good you did despite that nonsense,” Therriault wrote.
Therriault went on to say in another deleted tweet that models he’d seen “never had mi/wi/pa looking even close to safe. Her team thought they knew better.”
Reached for further comment on Thursday, Therriault told The Daily Beast: “Thanks, but I'm not interested in talking further at this time.”
In many of the post-mortems on the failed Democratic bid, Clinton’s campaign was accused of missing or simply not believing warning signs in the so-called Blue Wall, midwestern states like Wisconsin and Michigan that were reliably Democratic strongholds but went to Trump last year. This, in the minds of data gurus is an error in judgment from the campaign, not an error in the actual data that was available to them.
“There was absolutely a failure by the campaign,” Tom Bonier, the CEO of TargetSmart, told The Daily Beast. “That had nothing to do with the data they got. It was about the way they used that data.”
Bonier said that he was “honestly shocked” when he heard the comments from Clinton and said they were “frankly wrongheaded.”
“The reason why it was so surprising from a perspective of veracity was that it was so contrary to the truth,” he said.
However, people like Bonier were quick to admit that electoral data is far from an exact science and that in the last month or so of the campaign, Trump got every lucky break in the book.
“We were surprised by the surge in turnout among white working class voters,” Bonier said. “We were surprised by the decline in turnout by African Americans. All of these other surprises added up. That’s not to blame the campaign for not seeing it.”
But still, Bonier said that the confidence of the Clinton campaign, which often led them to assume that reliably blue states would go their way, ended up being a bit of an achilles heel.
“It was a failure of imagination to not try to figure out ‘Is there a scenario where Trump could win?’” Bonier said.
The question now for many in the field of data analysis and voter turnout predictions is how to move forward after such an unprecedented election result.
“We will likely consider the implications of being wrong much more frequently than we have in the past,” Bonier said. “And consider ‘Am I being overly optimistic in considering different scenarios?’”
One specific issue Hagner noticed in 2016, according to surveys his company conducted, was that Clinton voters answered far more questions about their voting preferences than Trump voters did in many circumstances.
“People who were going to vote for Hillary would tell you all the information you’d want to know,” Hagner told The Daily Beast. While some inclined to vote for Trump would hang up without explaining their decision-making process further.
“If you need to understand where the race stands, there are just voters who won’t take part,” he said describing the issue.
The position of those in the field of Democratic data is perhaps best summed up in a piece that Therriault wrote in February.
“Data is a tool for campaigns, not a strategy,” he said.
And for most people, even the Democratic nominee, it can’t be an excuse either.